In order to test the possible cardiac-sparing effect of doxorubicin administered by six-hour intravenous infusion and to prospectively evaluate the role of resting left ventricular ejection fraction in monitoring these patients, 33 women with advanced breast cancer were treated with combination chemotherapy containing 5-fluorouracil, cyclophosphamide, and doxorubicin. Doxorubicin was administered via a femoral catheter as a six-hour infusion. Cardiac function was monitored prior to therapy and at intervals during therapy by history and physical examination and by measurement of resting left ventricular ejection fraction with gated pool radionuclide angiography. Twenty-six responses were observed (complete response, seven [21 percent]; partial response, 19 [57 percent]). Systemic toxicity included alopecia, myelosuppression, and nausea and vomiting. There was a progressive fall in resting left ventricular ejection fraction during treatment from a median baseline value of 0.63. Mean fall from baseline left ventricular ejection fraction at a cumulative doxorubicin dose of 200 to 300 mg/m2 was 0.06 (p less than 0.005); at 301 to 449 mg/m2 it was 0.09 (p less than 0.0005); and at 450 mg/m2 or greater it was 0.15 (p less than 0.0005). Clinical congestive heart failure developed in three patients. Even though the decrease in left ventricular ejection fraction was often within the "normal range" (left ventricular ejection fraction 0.50 or greater), these changes were progressive and appeared to be part of a continuum of doxorubicin-induced myocardial damage. Steady-state infusion levels of doxorubicin in plasma ranged from 90 to 120 nM. They confirm the hypothesis that lower concentrations can be achieved by continuous infusion rather than by bolus infusion. In this study, however, administration of doxorubicin by six-hour infusion did not appear to have a major cardiac-sparing effect. Studies of anthracycline cardiac toxicity should include determination of baseline left ventricular ejection fraction and serial observations during therapy. Failure to include deteriorations in function above an arbitrary cutoff point or to make observations only at higher cumulative doses may underestimate drug-induced myocardial damage.